The present invention relates generally to optical sensor systems and more particularly to an optical sensor system capable of distinguishing between different illuminants.
The human vision system is a very poorly understood mechanism that translates photons of various wavelengths into visual pictures that human brains can understand and respond to. The human vision system and mental system compensate for scenes under various illumination sources and provides to the viewer a xe2x80x9ccorrectedxe2x80x9d visual picture. For example, white tee shirts appear white in human vision regardless of whether the scene happened under noonday sunlight or in the last minutes of a red sunset. When digital cameras, either video motion cameras (VMC) or a digital still camera (DSC), are exposed to similar illumination environments, the resulting images are profoundly different.
Extensive research has been undertaken to predict a mathematical construct for an image called the White Point (WP). The WP is the illumination that occurred at the brightest point in the image and represents what should be considered xe2x80x9cwhitexe2x80x9d in the final image. It is assumed that every image has some white objects or highlights in it. When the brightest object, with roughly equal amounts of red, green, and blue is located, the WP operation is constructed by determining the multipliers of the red, green, and blue parts of the brightest point so that the resulting red, green, and blue values will be made equal. Once this transformation is known for the brightest point in an image, it is simultaneously applied to all the other points (which are called pixels) in the image. The WP operation typically results in a final image that looks much more realistic with respect to its color balance.
There is a significant shortcoming of the simplistic WP operation described above. It is the assumption that there are some spectrally xe2x80x9cwhitexe2x80x9d objects in the image. While this is true the majority of the time for typical xe2x80x9ctouristxe2x80x9d pictures, there are also numerous cases where a spectrally xe2x80x9cwhitexe2x80x9d object is not present. For example, a close-up picture of a red barn with some blue and green metal signs attached to the barn""s side. The dominant color would be red. Some digital camera systems might interpret the large amount of red as a color cast problem that typically occurs under sunset illumination conditions. The brightest part of the image would be the green signs. If the digital camera algorithm attempted to use the green area as the WIP, then the resulting picture would be made very blue. The underlying problem is caused by not knowing the true nature of the illumination (light source) present at the time the image was captured.
Current technology tries to use the color content of the image to estimate the color illumination. In video motion cameras (VMCs), there has been remarkable success with this method since the videographer typically xe2x80x9cpansxe2x80x9d a scene to cover a large area. In this process, there is almost always some bright white object that can be identified in the multiple images. Once this xe2x80x9cbrightestxe2x80x9d object is imaged, the WP algorithm locks in on this area and makes a prudent estimate of the white point illumination and keeps this WP value until a xe2x80x9cbrighterxe2x80x9d white object is discovered. This is not true for digital still cameras (DSCs) where typically only a single image is capture for a given scene.
A great deal of research is being conducted to see if the WVv of an image can be deduced from just the image itself. However, examples like the barn picture described above will always cause problems. An alternative solution is to measure the scene""s illumination source directly. In black and white photography, the measurement was performed with a xe2x80x9clight meterxe2x80x9d. The meter is pointed at the light source, which would be straight up for daylight or towards a spotlight if it were focused on the object of interest. In color photography, a more sophisticated type of xe2x80x9clight meterxe2x80x9d called a photo spectroradiometer is used. Rather than measuring a single quantity like the black and white light meter, a photo spectroradiometer has to measure numerous points across the visual light spectrum and make a graph of the power at each wavelength that it has found. Once this graph is known, then an accurate representation of the original image can be constructed by removing the influence of the light source from the original scene. For example, an image of a white tee shirt at sunset will have a definite red cast to it. The photo spectroradiometer graph will show strong photon power in the red region of the visible spectrum. Knowing how much influence the illumination source had on the resulting image, a mathematical process is performed to remove the dominant red from the image. The final image has the white tee shirt looking truly white. In the other example of the red barn with the blue and green signs, the photo spectroradiometer graph would show normal daylight present as the illuminant. This means that almost no color correction would be applied to the final image. So in this case the dominant red barn color would be left in the image since that is the normal color that human vision would have seen under midday circumstances. The photo spectroradiometer is the ideal instrument for taking color pictures.
The problem is that a spectroradiometer is both big and expensive. A typical unit is 10 by 6 by 4 inches in size and costs between $5000 to $50,000 in 1998 dollars. It also requires a computer to readout its graphical data and apply it to the image in question. What has long been needed is a low-cost, small, portable spectroradiometer to indicate the type of illumination present while a picture is being captured.
The high cost of spectroradiometers comes from the narrow bandwidth samples (typically ten nanometers in width) that they provide. Narrow bandwidth measurements are essential for scientific calculations, but while working on the present invention it has been determined that they are not required for illumination discrimination.
The present invention provides an optical sensor system which uses portions of the intensity spectrums of various types of natural and artificial light and combinations thereof to determine the nature of the illuminants. The intensity spectrums are sensed by a plurality of photosensors connecting to a processing system which can discriminate characteristic areas therein. The present invention measures relative mixtures of sunlight, and artificial light, such as tungsten, fluorescent, and xenon photoflash.
The present invention further provides a plurality of photosensors having a plurality of bandpass filters which allow discrimination of the sources of illumination, such as natural light, artificial light, and a combination thereof.
The present invention further provides a plurality of photosensors having a plurality of bandpass filters selected to optimize the discrimination ability for various types of illuminants.
The present invention further provides a minimum of five bandpass filters measurements of the spectral power that can resolve daylight, incandescent (tungsten), and florescent light sources.
The present invention further provides a plurality of photosensors having a diffraction grating which allows discrimination of the sources of illumination, such as natural light, artificial light, and a combination thereof.
The present invention further provides a plurality of photosensors having a diffraction grating selected to optimize the discrimination ability for various types of illuminants.
The present invention further provides a computer program for selecting bandpass filters to optimize the discrimination ability thereof.
The present invention further provides a computer program for selecting dye-based bandpass filters with maximum discrimination ability for minimal cost.
The above and additional advantages of the present invention will become apparent to those skilled in the art from a reading of the following detailed description when taken in conjunction with the accompanying drawings.